Identification of models for predictive control

In this paper, the assumption is made that one needs a model on which to base the design of a predictive control law. In order to get the best performance out of the controller, there needs to be a synergy between the model identification algorithm and the control design so that they are mutually supportive. Here it is demonstrated how a particular, rather unnoticed approach, can produce suprisingly good results compared to the more usual techniques adopted. The method uses models specifically identified and designed for a given prediction horizon